There are many discusions about monitor calibration, which colorimeter to use, which software, colour temp setting etc...

But after you have calibrated your monitor using lets say an Optix XR, how do you actualy verify if the monitor colours are right?

...to go one step further (I intend to buy an new monitor) how do you judge the colour capabilities of a monitor? I know that Gretag Macbeth sells colour check cards, but can you use those cards to actualy verify if the monitor is well calibrated and which colour spectrum it can display...or is this a stupid question..

I am not a pro so I just ask...

I don't think there's any way to visually verify the success of the calibration. Experts could probably detect a problem, but that's an acquired skill.

I also don't think comparing a Gretag card is going to help much either. The difference in the quality of light between the reflected light of the card and transmitted light of the monitor make such a comparison very difficult for us mere mortals.

Any calibration technique incorporating subjective assessment (eg Adobe Gamma) is fraught with hazzards.

There are a variety of techniques to use. I strongly recommend getting a Gretag Colorchecker chart, it is a reference that once also uses for print comparison. Then you compare the screen image of the chart with your physical chart and this gives you an insight into the quality of calibration and the quality of your monitor.

Also, you need to print some sample imagery *in your favorite syle* on a profiled printer and check against what the monitor is showing - after all this is the purpose of this game.

Another test you should do is create a document in the monitor working space (yes, really), and then create a grey scale gradient in it. This should show you how the shadow and highligts are displaying, and how neutral the profile is. Good screen profiles show both ends, and are smooth, with no bands or reversals.

Edmund

To save my lazy self some typing, here's Andrew Rodney's article on the subject:
http://www.ppmag.com/reviews/200412_rodneycm.pdf


That technique introduces quite a few variables (particularly room lighting) unnecessarily. Testing for blacks and artifacts (refer to linked article) are more effective, not to mention easier since you deal directly with the profile and are not depending upon the subjectivity of your eyes.

Okay but if I wan't to compare a colorchecker chart you also need the digital version of the chart to correctly verify what's on your screen versus you chart? Or are these colour numbers available in Photoshop by default?

I did the tests described in the link to Andrew Rodney's article. I had a simillar test from NEC, but that one mainly focusses on the black level and if there are any artifacts showing up in a black-white gradient.


Thanks for posting, it's usfull info

PS I just found the GretagMacbeth colour charts in a digital form at: DryCreek

And that's what counts. If black levels are good and you have a smooth grayscale gradient, all else will render correctly. Looking at all the colors of a color chart will not net you any additional useful information.

Really ? Maybe people who subscribe to this idea just need black and white screens.

Edmund

Here is my LOW tech calibration method.

I use a one lab to output my critical work on Fuji Crystal Archive paper.

I have a portrait of a couple on my "desktop" that has been printed with no color corections by this lab. I open the file and reference it to an item I am going to have printed by this lab, and compare it to what is going to go. I know this sound confusing but it works for me for 4 years now and three computer set-ups. I find that my Sony Monitor (LCD) and Apple computers don't need much adjustment to make that reference photo "look right" as my benchmark.

I also don't seem to get many complaints if any from images I send for publication to dozens of offset printers each year. I am always "in the ballpark".

--for what it's worth

Just so I understand what you're saying...

I believe the question is predicated on determining the quality of the callibration AFTER you have run an Eye One or Gretag etc etc. I understand you to be saying that you can eyeball a printed Gretag chart against a displayed chart on a recently callibrated monitor and visually decide that the monitor is not acceptably close enough the the printed chart. Is that correct?

I did the checks after re-profiling today, and the monitor is decidedly NOT right. I used BasICColor 4.02 on the Eizo CE210W, and the result has a definite magenta cast. The profile validation shows some pretty ordinary results for the colour patches, and very poor results for the greys (mostly in the red).

Settings - D65, L*, target white 115cd, min black.

What does one do when this happens? I can drag points on the curves in the adjustment feature the software has, but it's coarse and depends on visual accuracy and a lot of trial and error.

Peter

I re-calibrated using 'Native' rather than D65, and the output looks visually more neutral, but still not right. BasICColor continually fails to hit my brightness target too - I set it for 115, it returned 125 cd. First time, I set it for 120, it returned 130.

The validation shows that the greys are better (out of the red, at least) but the colour patches fared worse, though they're still in the green zone.

Anyone got any bright ideas? Is this as good as I can expect from this monitor, or is there something I'm messing up?

That is, after all the purpose of the profiling - to match what you see on screen with what comes out of the printer. I visually check the monitor calibration with a couple of test prints plus a grey ramp / continuous tone, to look for any gross issues.

I'm certainly seeing noticeable colour and neutrality differences between this monitor profile and a well-calibrated print - ones which I didn't have before I changed the monitor.

Peter

I assume this is an LCD ?

The brightness differential is related to the high native color temperature of the display.
You should let the display warm up for half an hour before calibrating. Backlights fluctuate substantially when warming up.

If you are on a PC, I would recommend checking whether some old pieces of monitor have hooked in profile-loaders, and disable those. Also, I would try firing off a message to Basiccolor, and experiment with some other software eg. ColorEyes.

Maybe your calibration device is unhappy with you or the screen. And then, maybe, if this is a cheap LCD it may be cheap rather than good.

Edmund

Correct: You want to check the calibrated monitor against a Colorchecker chart. This chart is not printed, it is actually something more like painted, and unfortunately it's expensive. Putting up an image of this chart on the screen and comparing it to the original is one (just one) of the accepted ways of getting a "feeling" for the quality of the monitor and the calibration software.

Male color consultants, male painters, and much of the female population can actually discern color variations between different production samples of the Colorchecker chart. Consultants may measure the physical chart and substitute the values into the screen image; however this degree of obsession is probably not healthy

Edmund

Edmund, the display is indeed an LCD - it's the Eizo CE210W, definitely not a cheap monitor, marketed as geared towards colour accuracy: one step down from the CG range. Approximately US$1,200 worth.

Before calibration I ensured the screen and power savers were off, and it warmed up for about four hours whilst I was working at my day job.

This is a new monitor, and a new graphics card (Radeon X1600PRO dual DVI). I had 2x 20" Apple Cinema Displays before on an Asus card, one of the monitors died. I de-installed the PowerStrip drivers I created for the ACDs, I did not use WinACD. I also de-installed all the previous graphics card drivers before loading my new card, and there have never been any other monitors on this system - no possibility for rougue monitor drivers. I re-added the second, non-dead ACD monitor and it calibrated as well as it did before.

There are no other profile loaders in my PC, I can verify that the correct profile is loaded through the 'Color' WinXP control panel. I have no other monitor profiles loaded except the ones I create. Adobe Monitor Gamma (or whatever it's called) has not been on my system ever - that's not interfering.

I demo-tested ColorEyes before I bought BasICColor, I don't think I can demo it again. I'm loath to buy another calibration software unless I can be sure that it will sort out my issue. One thing in favour of ColorEyes was that I found Jack Bingham to be very helpful and responsive; I doubt I'll get anywhere with BasICColor over the weekend...

BTW, the calibration device is the DTP-94

Im am not a calibration expert but the CE210/CE240 are hardware calibrated that means the calibration is done inside the monitor using the Eizo software. If this is true I think their should be no profile active that affects the LUT in the videocard. The video card should have no profile or a neutral one. You could make the mistake that there is a calibration profile active inside the monitor, while another one is active for the videocard. Normaly the video cards LUT is used to calibrate the monitor, but the Eizo CE, CG and some NECS with spectraview software are calibrated inside the monitor. Maybe you should check the Eizo documentation on the subject for calibration and monitor profiling.

I wonder if in your case the black-white gradient looks neutral and without artifacts.

just my $0.02

Ok, you've eyeballed the callibrated monitor next to a gretag chart and decided there are differences that you'd rather not live with. In my view the skill required to accurately make that judgement is acquired only after __considerable__ experience, and is not likely a skill possess by one just starting out on the color management journey, but regardless.... What next?

I see only three options (none of which make particular sense to me) 1. Recallibrate with the same hardware 2. get a different hardware tool, 3. start to manually tweak based on what you see (which makes no sense - why would you bother with hardware callibration if you could get superior callibration manually) - hey - there's always adobe gamma

Having said that, I certainly understand that it's always possible that the hardware callibration device is grossly malfunctioning, and if that's the scenario we're dealing with then I don't have any issues.

FWIW I assume you'll agree that without a controlled light source with which to view the gretag chart you're wasting your time?

The gradient looks smooth, but not neutral - there is a pronounced magenta cast.

The BasICColor software takes account of the ability of the monitor to be hardware calibrated, and offers the option for Hardware Calibration (monitor LUTs), which I selected. I don't know how to check if there is a monitor profile active as well as a video card profile - I don't think so, but I'll do a bit more reading - the Eizo documentation is crap.

Cheers,
Peter

I found that using a Solux lamp enabled me to calibrate my DSLR, and also works great for verifying your CRT calibration.

I used this pic to check my calibration. Process your image twice, once with WB on the CRT gray patch, once on the print patch. Crop out the print in each image, stretch any which is not rectangular to make comparison easier, and eyeball them.

I found that my 6500K crt WB at 6600K, and my Solux (which measures 4175K with an EyeOne Pro spectro) Wb at 5250K (both in ACR). There are very slight differences in just a couple patches.

When I was first using two ACD monitors, I was reasonably happy with the calibration I got from the Monaco Optix Pro software. I never got perfect match to my prints, but then I had a low-grade printer so I didn't expect a lot more. Later, I upgraded the printer and Mr Rodney made me an excellent profile, which enabled me to see that my monitor calibration was off. I bought BasICColor which gave me profiles that were a far better match to the print, and I finally had excellent monitor to print matching.

I replaced a dead ACD with the Eizo CE210W which meant I also had to replace my video card, so all was sent back to square one. Initially I calibrated the Eizo with it's own software, and it looked fine to me, but the ACD looked quite different - just a matter of re-doing the Apple, I thought. In the meantime I hadn't used the monitor for photos, I've just been working at my day job. When I finally got round to evaluating it properly, I could see that my initial non-critical impression was quite far off. The ACD was actually still pretty good using its old profile, but the Eizo was not good.

I've since manually adjusted the profile from within the BasICColor software and it now is a closer match, but not good enough. I'm struggling to know where I go from here.

BTW, I examine my prints under controlled 6500k lighting

Is there a way to return to the factory defaults, to see if the magenta cast dissapears? This way you can rule out if the color cast is caused by the calibration proces or not. I couldn't find any usfull info on this subject on the Eizo website or in the documentation (I downloaded the Colournavigator CE software).

René, I tried your advice by removing all profiles associated with the monitor, but it just seems to have retained the last profile. I'll dig into it a little more to see if there's some way I can do that, thanks for helping!

Peter

It's a pitty that until now, no one with real life experience in hardware calibratable monitors has responded yet. When using a hardware calibratable monitor like the Eizo CE210/240, CGxxx and the NEC's 2190UXi (spectraview) etc... the monitors internal LUT is adjusted while the graphicsboard LUT remains linear (neutral). But I don't know if the Eizo software, after calibrating the monitor automatically resets the graphicscard LUT and updates the profile (WIn2000 > Controlpanel->Display->Color management->Default Monitor profile).

I hope you can find the problem.. ..good luck.

Actually, I own a CG210 which has the writable hardware LUTs and DDC ability; I also have a cheap but decent Samsung monitor attached to the same Mac, and matched calibrations. I believe I am reasonably familiar with the issues involved, although I avoid using PCs.

If an Eizo failed to calibrate, I think

- I would first get rid of the other monitor, turning it off and disconnecting it from the system.
- I'd check to see whether the Eizo is actually in a calibratable mode and not in a hard-wired mode (front panel settings).
- I would then shift the Eizo to the primary head of the display card.
- I would delete any existing previously created profiles from the system.
- I would fire up Color Navigator or the latest version of the Optix software. If these two fail to give decent (not necessarily perfect) results when the screen is acting as a primary monitor, then I would diagnose a system or hardware problem (the puck may be broken). If they worked, then I would start looking for software that gives perfect results, and/or allows me to calibrate both screens - a much harder problem.

If I may politely says so, the best solution for someone with a system above a certain complexity is to call in a color consultant for an hour to fix it.

Edmund

Thanks René - unfortunately I have to travel on business so can't get back at this for a few days - I'll try to shoot an email to Eizo in the meantime to see if they can offer any advice or at least better documentation..

Edmund, thank you for your suggestions - I will try again to calibrate the Eizo after removing the ACD.

I believe my monitor IS in calibratable mode, as after applying the profile the display changes, and the colour settings are dimmed and no longer selectable from the front panel. The Eizo is already and always has been the primary monitor, and before I started any calibration I already removed (de-installed and deleted) all previous profiles.

The puzzler for me is that the validation SHOWS how poor this profile is, so it seems the puck is working fine (and the ACD re-calibration was OK).

I have to leave this now for several days as I have to travel; in the meantime I'll send some mails off to Eizo and BasICColor.

Peter

peter,

try use these setting(not express mode)

gamma L*
6500k
100cd/m2
16bit LUT base profile,CAT02 mode,ICC V4 profile

remember,display type is "LCD"
calibration method must use"software(video LUT)"

and set your monitor in "custom mode", follow the popup adjust RGB and brightness value.good luck.

Yes, be sure to set the front panel buttons/menus to "custom mode". I hae a feeling your monitor may be set into one of the pre-calibrated modes.

Edmund

It seems to me that any method used to check the calibration would be a good method to calibrate your monitor. At some point, you just have to trust.

Well, I'm back from my travels and got to spend a hour on this darn monitor calibration. I now have two new profiles which are a lot better than before, and I got them by plugging the puck into a different USB socket! I made a software calibration as suggested (thanks Ato) and that came out very well indeed - the hardware calibration was not quite as good, but very close and definitely acceptable. Validation charts attached below: the first is the original result, the second is the hardware calibration in a different USB port, the last is the software calibration in the different USB port.

Click to view attachment Click to view attachment Click to view attachment

In searching the net for some clues about this behaviour, someone suggested that you shouldn't use the USB ports in the side of the monitor for its calibration - said he got better results by plugging the puck directly into the computer. Though I couldn't see what difference that would make I followed the advice and it worked!!

So the thing that's puzzling me now, is why on earth would a software calibration be better than the hardware calibration? I thought hardware calibration was the holy grail, and that's one of the reasons I bought this monitor.

And the profiles made by Eizo's ColorNavigator were all crap no matter what port I plugged into.

Thanks to those of you who took the trouble to reply, I'm grateful for your advice.

The main difference between hardware and software calbration is that when using hardware calibration the total span/range from each single graphicscard's RGB channel is fully used for displaying (selecting) colors.

This means that in the case of hardware calibration the video card can output R:0..255, G:0..255, B:0..255. The RGB adjustments are made inside the monitor in the monitors LUT. THis means the graphics adapter can select 256*256*256=16.8mlj colors.

In case of software calibration the videocards RGB output is manipulated to achieve the same correction. This could mean for example that the effective videocard output is (fictive values): R:6..230, B:23..244, G:10..255. This leads to less selectable colors: 224*221*245=12,1mlj colors.

If you have a display that is almost perfectly calibrated by the factory, the difference will be much less, compared with a bad display that needs a significant amount of compensation.


Windows Vista will support more then 8 bit RGB output. The ATI Fire GL cards are currently the only cards that support 10bit DVI (RGB=1024*1024*1024=1bilj colors), but there is only one monitor that supports it (as far as I now..the NEC 2180 Spectraview Reference). I have this information because I asked NEC if the NEC2690 would support more than 8 bit DVI, it doesn't.


You say the ColorNavigator software is crap, but what software did you use?

Not sure I understand the question - ColorNavigator CE is the calibration software: I used version 1.1.3, having made sure BasICColor was di-installed beforehand.

After re-reading you reply I see you did both a software calibration and hardware calibration using different USB ports and software.

On the GretegMacbeth website I found a tool that allows you to examine the curent videocards LUT settings and to reset, save and restore them.

GretegMacbeth Calibration checker THis way it's easy to check if your videocard is modifing the LUT (it has a profile loaded).

René, I downloaded that software and using it I discover that there is a video card LUT setting active even though my monitor is hardware calibrated - looks like there are two calibration settings being applied. I'm guessing that the videocard setting was in place whilst I did the hardware calibration of my monitor, and it adjusted for the non-linear values coming from the video card. I'm going to re-calibrate this morning after re-setting the videocard LUT, and see if that gets me a better result.

THANK YOU René!

By the way, seems my experiences with the USB port aren't so crazy - I'm hearing of more instances where using a USB hub creates calibration problems.

Ok this brings light in the tunnel. This small software package also allows you to reset the graphicscard LUT, so it's absolutely neutral. Then calibrate your screen again

I'll definitely do that. As luck would have it, my motherboard died and I've only just got back online again after half a day troubleshooting and eliminating all other possibilities first (my motherboard is 4 years old, none of my components - new videocard included - will work with a new one). I've got an old temp board in place for now, will get back to calibration once I re-build the PC!